Basics of Remote Sensing Some literature references Franklin, SE 2001 Remote Sensing for Sustainable Forest Management Lewis Publishers 407p Lillesand, Kiefer 2000 Remote Sensing and Image Interpretation Wiley 724p Wilkie, Finn 1996 Remote Sensing Imagery for Natural Resources Monitoring Columbia University Press New York 295p Schowengerdt, RA 1997 Remote Sensing Models and Methods for Image Processing Academic Press 522p Definition of Remote Sensing: Measuring without direct contact to the object to be measured Here: focus on digital imagery: digital representations of the electromagnetic radiation reflected (or emitted) from objects on the earth s surface Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 1 Fields of application Forestry Agriculture Environmental Monitoring Geology, geography Archeology Surveying, urban development, rural development Landscape planning Military wherever maps and / or other spatially explicit information is required Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 2 Applications in forestry Examples of uses of remote sensing imagery: Orientation in the field Recording of forestry relevant information, above all area information (forest,forest types,forest, fires, pests, damages) Observation of status quo and changes of areas and characteristics of areas Thematic maps and cartographic processing of forest inventory results BUT: Not everything that is of interest in forestry and forest inventory can be observed by means of remote sensing Therefore: For each single application we must find the optimal integration of different data and information sources Remote sensing is one among several Cost is a relevant criterion Modelling (=relating remote sensing visible features to ground features) Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 3 RS 1 p 1
Basic principles of remote sensing systems Remote sensing sensors register electromagnetic radiation (EMR) in one of the following forms: - reflected sunlight, - emitted radiation (all objects with Passive sensors temperatures above the absolute zero emit thermal radiation), - reflected microwaves (Radar) Active sensors or reflected infrarded (Lidar) Important for all applications are technical developments (new sensors, platforms), methodological developments (algorithms) Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 4 Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 5 Reflection characteristics of water, soils and vegetation Water: clear water appears blue as mainly short wave lengths are reflected (04-05 µm) Water absorbs completely the near IR (08-3 µm), so that it appears black there Depending on the suspension and other color giving features in the water it can appear differently For aspects of water quality assessment, sediment etc we would use sensors in the range of 04-06 µm To separate water from land: near IR 08-3 µm is better Problem: Atmospheric disturbance of radiation through water vapor Particularly important in the blue range 04-05 µm but also for green 05-06 µm Water vapor (and little drops = clouds) in the atmposhere inhibit the sight in some spectral ranges Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 6 RS 1 p 2
Soils and rocks: Soils are background in many vegetation covered areas and (together with rocks and artificial objects) the only reflecting objects in vegetation free areas Strong reflection in the visible spectrum The actual reflection characteristics depend mainly on soil moisture Soil reflection increases continuously from the blue wave length onwards over a broad range of wave lengths If multispectral images are available this characteristic (among others) can be used for the identification of soils Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 7 Typical spectral response of sandy soil of different moisture contents Moisture % Spectral reflection % Wave length Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 8 Vegetation: Vegetation is green (chlorophyll), absorbing particularly in the blue (mainly 041, 043 and 0453 µm) and in the red (mainly 0642, 0662 µm ) 047µm is the maximum of solar energy Main reflection and transmission is at green We tend to interpret green objects automatically as vegetation The cell structure of the leaves causes a strong reflection also in the near IR (around 08-11 µm ) Water contents of leaves causes absorption around 14 and 2 µm Vegetation absorbs (blue, red) and reflects (green, near IR) very typically, so that vegetation can often easily be distinguished from non-vegetation Variations in the reflection pattern indicate different pigment- or water content This helps to better identify species, species mix, health status, age, development class etc Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 9 RS 1 p 3
Reflection at green leaves SUN Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 10 Typical spectral response of leaves and the causes Pigments Cell structure Water contents of leaves Major cause Spectral reflection % Chlorophyll Water Absorption bands Wave length Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 11 Digital imagery consists of pixels (picture elements) : The radiation that is registered by the sensor is converted to digital numbers that correspond to the integrated radiation energy found there Within-pixel details are lost and can only in part be recovered (sub-pixel analysis) Upper left: Original digital image containing 320 rows and 480 columns Upper right: Amplification Lower right: Digital numbers for the amplified image Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 12 RS 1 p 4
Basic operating and recording characteristics of sensors for a pixel-wise registration of digital imagery Image-wise pixel-wise strip-of-pixel - wise Landsat MSS: 6 strips simultaneously Whiskbroom Pushbroom Source: Wilkie & Finn 1996 Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 13 Multi-spectral sensors: energy of radiance is separately registered in different spectral bands White light Energy of EMR is converted into electric energy (voltage) the intensity of which is registered for each spectral band Different spectral bands are separately recorded and can be analysed separately (difference to a panchromatic recording system) This allows more easily to identify typical reflection patterns and to produce color images Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 14 Landsat TM Spectral bands of remote sensing sensors coincide with the atmospheric transmission windows Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 15 RS 1 p 5
Multiple band digital images are huge amounts of data in three dimensions: - lines (y-coordinate) - columns (samples, x-coordinate) - bands (spectral bands) Common storage sequences in the data files: BSQ: Band sequential (sequence of complete images per band) BIS: Band interleaved by sample (pixel after pixel all bands) BIL: Band interleaved by line Example of a 8x8 pixel image with 7 spectral bands Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 16 Characteristics of sensors and digitally recorded images 1 Spatial resolution: Size of the smallest object that can be distinguished (pixel size) 2 Radiometric resolution: contrast; sensitivity in which radiation intensity is measured in the spectral bands = number of levels of gray values (stored in 6 bit 8 bit 10 bit format) 3 Spectral resolution : color; number and width of the spectral bands, and total spectral range covered 4 Temporal resolution : Frequency of recordings of the same location (for change assessment and construction of time series) Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 17 Spatial resolution Integration of radiance takes place over an ground area defined by the groundprojected instantaneous field of view (GIFOV) The interval at which those takings are made (= pixel locations) is called ground sample interval (GSI) Frequently but not necessarily: GIFOV = GSI Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 18 RS 1 p 6
File sizes? Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 19 Example Spectral Resolution Landsat 7: Sensor ETM+ = Enhanced Thematic Mapper Spectral Bands: Landsat 7 Visible/Near IR 45-52, 52-60, 63-69 76-90 Shortwav IR 155-175, 208-235 Thermal IR Panchomatic 5-9 104-125 (hight and low gain) Spatial resolution: 30m (thermal 60m, pan 15m) Radiometric resolution: 2 8 =256 intensity levels Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 20 1 2 3 4 5 6 7 Source: Wilkie & Finn 1996 Information contained in bands 1-7 of Landsat ETM+: Intensities of gray values in each band for a landscape To make it a color image, we must assign a color to each band: According to color theory any color is composed of components of Red/Green/Blue (RGB) Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 21 RS 1 p 7
The resulting image is called Color Composite Only three bands can be depicted simultaneously (RGB) We must select those three bands that contain most information or that produce the desired repsentation Example: Band color assignments for satellite imagery that mimic true color (TC) or Color Infra Red (CIR) images: Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 22 Steps in a remote sensing application/project (after Wilkie & Finn 1996) 1 Formulate the problem 2 Obtain data (digitize maps or buy digital imagery) 3 Choose software image processing system (soft-, hardware) 4 Assess data quality (descriptive statistics, image display) 5 Correct errors: radiometric (atmospheric or sensor), and geometric 6 Enhance images (for digital or for visual analysis) 7 Conduct field survey 8 Extract features from imagery (classification, accuracy assessment 9 Input into a GIS 10 Summarize results Lecture to Remote Sensing, Georg-August-Universität Göttingen Slide No 23 RS 1 p 8